Starship Asterisk*

Knistian wrote: ↑Mon Apr 25, 2022 5:33 pm Regarding the resolution of the engineering-picture from JWST, where several think it's low: It has to do with wave-length. It's unfair to compare a picture taken in the UV-Visible range(where Hubbel opperates), and compare it directly with an image taken in the deeep part of the IR. The diameter of JWST is ca. 2.3 of Hubbel, so wavelengths 2.3 times as long as visible would be at the same resolution (with teoretical best acheivable resulution from both). The Webb has acheived to get a resolution at the wavelength used in the image, that is not limited by missalignement by the optics, but by fundamental limitations given by the diameter of the primaty mirror.

Knistian wrote: ↑Mon Apr 25, 2022 5:33 pm

AVAO wrote: ↑Sat Mar 19, 2022 4:34 pm I'm a bit surprised that the quality isn't better than Hubble when comparing the resolution of details to the nearest Hubble object. The third image in the series shows the nearby Cat's Eye Nebula overlaid at the same size. But of course, the process of optimizing the instruments is not yet complete.
...

Regarding the resolution of the engineering-picture from JWST, where several think it's low: It has to do with wave-length. It's unfair to compare a picture taken in the UV-Visible range(where Hubbel opperates), and compare it directly with an image taken in the deeep part of the IR. The diameter of JWST is ca. 2.3 of Hubbel, so wavelengths 2.3 times as long as visible would be at the same resolution (with teoretical best acheivable resulution from both). The Webb has acheived to get a resolution at the wavelength used in the image, that is not limited by missalignement by the optics, but by fundamental limitations given by the diameter of the primaty mirror.

AVAO wrote: ↑Sat Mar 19, 2022 4:34 pm I'm a bit surprised that the quality isn't better than Hubble when comparing the resolution of details to the nearest Hubble object. The third image in the series shows the nearby Cat's Eye Nebula overlaid at the same size. But of course, the process of optimizing the instruments is not yet complete.


https://live.staticflickr.com/65535/519 ... 2763_k.jpg
https://sky.esa.int/?target=268.9184271 ... ue&lang=en

https://live.staticflickr.com/65535/519 ... 412c_k.jpg

https://live.staticflickr.com/65535/519 ... fc11_k.jpg
Jac Berne (flickr)

Chris Peterson wrote: ↑Sun Mar 20, 2022 4:40 pm

johnnydeep wrote: ↑Sun Mar 20, 2022 4:05 pm So, what explains the 8 diffraction spikes we see in this image? There are the six large equal spaced ones, which I assume are due to the hexagonal mirror segments, but there are also two small horizontal ones. Or are those not diffraction spikes at all and just "processing artifacts" as I thought might be being hinted at in some of the posts above? Also, where are the diffraction spikes caused by the three secondary mirror support struts?

Also, I notice that the diffraction spikes exhibit some striated complexity - what's causing that?

The secondary mirror is supported by two struts that are aligned with the hexagonal mirror vertices, and therefore their diffraction spikes overlap those from the mirrors themselves. The upper support strut, however, is aligned with the mirror edges and its diffraction pattern is seen in the much less prominent horizontal spikes.

Not sure about the striation. Might be related to the massive overexposure of this image and the effects of the internal mirror edges. Might try playing around with the math a little later.

UPDATE: Here's the aperture pattern (mirror segments and support struts), and the Fourier transform of that pattern, which is basically what the diffraction pattern is. (Zoom in for a better view.)
_
JWST_optics_fourier.jpg

johnnydeep wrote: ↑Sun Mar 20, 2022 4:05 pm So, what explains the 8 diffraction spikes we see in this image? There are the six large equal spaced ones, which I assume are due to the hexagonal mirror segments, but there are also two small horizontal ones. Or are those not diffraction spikes at all and just "processing artifacts" as I thought might be being hinted at in some of the posts above? Also, where are the diffraction spikes caused by the three secondary mirror support struts?

Also, I notice that the diffraction spikes exhibit some striated complexity - what's causing that?

Chris Peterson wrote: ↑Sat Mar 19, 2022 2:26 pm

JohnD wrote: ↑Sat Mar 19, 2022 2:17 pm The blurb for this APoD says, "The resulting image taken by Webb's NIRcam demonstrates their precise alignment is the best physics will allow."
The link in that sentence takes you to a previous APoD that describes "adaptive optics" on an Earthbound telescope, that is just not relevant for the Webb, out in Space!

Actually, it is very relevant. The JWST uses the same sorts of adaptive optics found on ground-based telescopes. That is, a closed-loop system that performs wavefront analysis and dynamically adjusts the position of optics in the system to achieve the highest performance. This requires faster corrections on Earth because of the atmosphere, but the approach is substantially the same, regardless of the actual sources of wavefront errors.

Chris Peterson wrote: ↑Sat Mar 19, 2022 6:45 pm

Ann wrote: ↑Sat Mar 19, 2022 6:29 pm

AVAO wrote: ↑Sat Mar 19, 2022 4:34 pm I'm a bit surprised that the quality isn't better than Hubble when comparing the resolution of details to the nearest Hubble object. The third image in the series shows the nearby Cat's Eye Nebula overlaid at the same size. But of course, the process of optimizing the instruments is not yet complete.

I thought the same, and I was unimpressed. But like you said, the process of optimizing the instruments is indeed not yet complete.
Ann

You're being fooled by the fact that most of the objects in the image are substantially saturated, which hides any detail. Look at the galaxies that are dim enough to not saturate. They show much finer detail than we see in the overlaid nebula. This image is already higher resolution than HST can produce.
_
51947342296_b9a8bbfc11_kx2.jpg

Ann wrote: ↑Sat Mar 19, 2022 6:29 pm

AVAO wrote: ↑Sat Mar 19, 2022 4:34 pm I'm a bit surprised that the quality isn't better than Hubble when comparing the resolution of details to the nearest Hubble object. The third image in the series shows the nearby Cat's Eye Nebula overlaid at the same size. But of course, the process of optimizing the instruments is not yet complete.


https://live.staticflickr.com/65535/519 ... 2763_k.jpg
https://sky.esa.int/?target=268.9184271 ... ue&lang=en

https://live.staticflickr.com/65535/519 ... 412c_k.jpg

https://live.staticflickr.com/65535/519 ... fc11_k.jpg
Jac Berne (flickr)

I thought the same, and I was unimpressed. But like you said, the process of optimizing the instruments is indeed not yet complete.

Ann

AVAO wrote: ↑Sat Mar 19, 2022 4:34 pm I'm a bit surprised that the quality isn't better than Hubble when comparing the resolution of details to the nearest Hubble object. The third image in the series shows the nearby Cat's Eye Nebula overlaid at the same size. But of course, the process of optimizing the instruments is not yet complete.


https://live.staticflickr.com/65535/519 ... 2763_k.jpg
https://sky.esa.int/?target=268.9184271 ... ue&lang=en

https://live.staticflickr.com/65535/519 ... 412c_k.jpg

https://live.staticflickr.com/65535/519 ... fc11_k.jpg
Jac Berne (flickr)

Ace Goodwin wrote: ↑Sat Mar 19, 2022 5:39 pm Those are some nasty diffraction spikes, and I don't believe getting used to them is the right answer, so... I would instead like to propose using new machine learning methods to eliminate them (and perhaps even the Airy disk as well) from images.
It goes something like this:
Create a synthetic data set by computationally simulating point source stars as viewed through an optical system (the intricacies of the point spread function tax my very limited math skills far beyond the breaking point sadly). Strictly speaking with some planning and the right equipment one could use a real point source such as a laser and a real optical system to achieve a training set tuned specifically to that optical system as well.
Using the original points image as the target and the simulated 'scope image as input, train a conv U-net or GAN (or any other relevant model).
If it's done right it should be able to essentially 'learn' to correctly reintegrate light from the spike regions back into the point sources from which they came. Then it's just a matter of doing model inference on real images.
I might be naive in thinking this can be done, but ml has already accomplished many amazing and "impossible" things, so I would love to take a crack at it (or even see someone far smarter than I do it). Any help with the math side of generating the data set would be appreciated.

AVAO wrote: ↑Sat Mar 19, 2022 4:34 pm I'm a bit surprised that the quality isn't better than Hubble when comparing the resolution of details to the nearest Hubble object. The third image in the series shows the nearby Cat's Eye Nebula overlaid at the same size. But of course, the process of optimizing the instruments is not yet complete.

Ann wrote: ↑Sat Mar 19, 2022 4:08 pm

Chris Peterson wrote: ↑Sat Mar 19, 2022 1:37 pm

Ann wrote: ↑Sat Mar 19, 2022 5:32 am
Thanks, alter-ego. However, I noticed that there was no spectral class given for 2MASS J17554042+6551277. As for the fluxes, the star appears to be ~1 magnitude brighter in V than in B, and it gets ever brighter at still longer wavelengths, but not enormously so. I would guess that this is a K-type star.

If you're feeling more ambitious, you can use a stellar flux calibration table (e.g. http://www.pas.rochester.edu/%7Eemamaje ... s_Teff.txt) and find the best match given ratios of different spectral bands. Just compute the ratios shown in the table using the SIMBAD data, and then look for the row that most closely matches those. You can probably narrow it down to stellar subclass.

Very interesting, Chris. Many of the rations suggest that 2MASS J17554042+6551277 is a star of spectral class K3V or K4V. However, that is impossible, since the star is 13 times brighter than the Sun in visual (V) light. No K3 or K4 dwarf star can do that. Therefore, 2MASS J17554042+6551277 must be a giant. It is clearly a very modest giant of, perhaps, spectral class K0III.

Chris Peterson wrote: ↑Sat Mar 19, 2022 1:37 pm

Ann wrote: ↑Sat Mar 19, 2022 5:32 am

alter-ego wrote: ↑Sat Mar 19, 2022 4:41 am m_v = 10.95
Clicking the 2MASS J17554042+6551277 link will take you to the SIMBAD data page for that star. There are several magnitudes for different spectral bands.

Thanks, alter-ego. However, I noticed that there was no spectral class given for 2MASS J17554042+6551277. As for the fluxes, the star appears to be ~1 magnitude brighter in V than in B, and it gets ever brighter at still longer wavelengths, but not enormously so. I would guess that this is a K-type star.

If you're feeling more ambitious, you can use a stellar flux calibration table (e.g. http://www.pas.rochester.edu/%7Eemamaje ... s_Teff.txt) and find the best match given ratios of different spectral bands. Just compute the ratios shown in the table using the SIMBAD data, and then look for the row that most closely matches those. You can probably narrow it down to stellar subclass.

skildude wrote: ↑Sat Mar 19, 2022 1:48 pm Is it me or is there a few false false diffraction lines on the photo? Say to the left of the star. Perhaps they took out stars that would mar the image. However, the diffraction lines on the bottom right clearly are not aligned. I also see that a couple of galaxies have lines moving away from the center of the image as well.

JohnD wrote: ↑Sat Mar 19, 2022 2:17 pm The blurb for this APoD says, "The resulting image taken by Webb's NIRcam demonstrates their precise alignment is the best physics will allow."
The link in that sentence takes you to a previous APoD that describes "adaptive optics" on an Earthbound telescope, that is just not relevant for the Webb, out in Space!

Ann wrote: ↑Sat Mar 19, 2022 5:32 am

alter-ego wrote: ↑Sat Mar 19, 2022 4:41 am m_v = 10.95
Clicking the 2MASS J17554042+6551277 link will take you to the SIMBAD data page for that star. There are several magnitudes for different spectral bands.

Thanks, alter-ego. However, I noticed that there was no spectral class given for 2MASS J17554042+6551277. As for the fluxes, the star appears to be ~1 magnitude brighter in V than in B, and it gets ever brighter at still longer wavelengths, but not enormously so. I would guess that this is a K-type star.